Clinical analysis of 73 children with anti-N-methyl-D-aspartic receptor (NMDAR) encephalitis

Introduction

Autoimmune encephalitis (AE) constitutes a unique form of encephalitis that occurs through autoimmune mechanisms, and is characterized by diffuse or multifocal inflammatory lesions in the brain parenchyma and resulting neurological deficits. Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis was initially described by Titulaer et al. (1) in 2013 in a cohort study of 577 patients, including 211 children, focusing on treatment outcomes and prognostic factors. The first pediatric case was reported by Höftberger et al. (2) in 2015, highlighting the clinical features and antibody profiles in children. The initiating triggers are varied and include viral infections and tumours (3), which cause the formation of autoantibodies to glutamate NMDARs, leading to an autoimmune response with both neurological and neuropsychiatric sequelae.

Epidemiological studies suggest that pediatric cases of anti-NMDAR encephalitis comprise a significant proportion of incidence, and that the disease is most prevalent among school-aged females (4,5). However, atypical presentation is commonplace for children, although this is especially notable in children <12 years, where the diagnostic process is often prolonged and delayed treatment is common.

Currently, anti-NMDAR immunoglobulin G (IgG) antibodies are assessed in cerebrospinal fluid (CSF) for diagnosis, often in conjunction with supportive imaging findings, biochemical, and radiological evaluations. Among the electroencephalography (EEG) changes observed in anti-NMDAR encephalitis, the Extreme Delta Brush (EDB) pattern has been proposed as a specific diagnostic test, as well as a potential indicator of clinical prognosis and treatment response (6). Cerebral magnetic resonance imaging (MRI) may reveal nonspecific T2/fluid-attenuated inversion recovery (FLAIR) hyperintensities, and it has been proposed that whole-brain atrophy carries a high correlation with poor outcome (7).

The Chinese medical community has established consensus guidelines for anti-NMDAR encephalitis treatment. Early immunotherapy leads to improved outcomes, but there is still a percentage of pediatric cases that either have poor status at follow-up or mortality (8).

In pediatric practice, decisions are often informed by early clinical prediction models of therapeutic response, using short-term outcome measures such as the modified Rankin Scale (mRS) at 2 months post-onset of encephalitis to assess functional outcomes. While children often demonstrate continued neurological recovery beyond two months, the 2-month mRS score serves as a critical early benchmark to evaluate initial immunotherapy responsiveness, and guide individualized rehabilitation planning—particularly for patients with persistent deficits (e.g., speech/motor impairments or cognitive dysfunction). This standardized timepoint facilitates multidisciplinary interventions, including physical/occupational therapy and neuropsychological support, tailored to residual disability levels (9).

Although prognostic markers are becoming increasingly recognised, in particular peak disease severity and imaging abnormalities, there are currently no validated pediatric-specific prognostic models. Current methods, such as the anti-NMDAR Encephalitis One-Year Functional Status (NEOS) score, which are primarily developed for use in an adult population, are often not representative of the clinical course in children. There is an urgent need to assess whether pediatric cohorts can adopt the existing prognostic model or if new models that take into account how children respond to immunotherapy need to be constructed (10,11).

The goal of this study is to systematically assess the clinical characteristics and treatment outcomes of pediatric anti-NMDAR encephalitis and identify independent risk factors that contribute to poor early prognostic outcomes. This study aims to investigate the clinical characteristics and identify prognostic risk factors in pediatric patients with anti-NMDAR encephalitis, thereby providing evidence-based insights for clinical diagnosis and treatment optimization. We present this article in accordance with the STROBE reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-401/rc).

Methods

Study population

A retrospective cohort study was conducted, enrolling 73 pediatric patients diagnosed with anti-NMDAR encephalitis at The Second Affiliated Hospital of Guangxi Medical University between January 2013 and December 2020. Comprehensive clinical data, including demographic characteristics (gender, age at onset), clinical manifestations, mRS scores at disease peak, time interval from symptom onset to diagnosis, CSF and serum anti-NMDAR antibody levels, neurophysiological and imaging findings [EEG, MRI, chest and abdominal computed tomography (CT)], treatment regimens, and outcomes, were collected through inpatient records, specialist outpatient visits, and telephone follow-ups.

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of The Second Affiliated Hospital of Guangxi Medical University [approval No. 2025-KYL(020)]. Informed consent was obtained from all patients’ legal guardians prior to participation.

Diagnostic criteria

The diagnostic criteria for anti-NMDAR encephalitis were applied based on the time of diagnosis: for cases diagnosed between 2013 and 2016, the diagnostic framework was based on the original criteria proposed by Dalmau et al. (2007) (11). For cases diagnosed between 2016 and 2020, the updated guidelines by Graus et al. (2016) (12) were applied. Although both criteria require the presence of characteristic clinical symptoms and antibody confirmation, the Graus et al. 2016 (12) criteria introduced more formalized clinical thresholds and broadened differential diagnostic considerations. As such, diagnostic consistency over time may have been affected. Patients were included if they met both: clinical criteria per Graus et al. 2016 (≥1 core symptom + exclusion of alternative); CSF-positive anti-NMDAR antibodies by cell-based assay (CBA) (Euroimmun, Lübeck) or Serum-positive antibodies only if CSF was unavailable (n=2, pre-2016 cases).

The 2020 International Consensus Recommendations for management of AE (13) emphasize: prompt initiation of first-line immunotherapy (steroids, IVIG, or plasmapheresis); early escalation to second-line agents (e.g., rituximab, cyclophosphamide) in cases of inadequate response; multidisciplinary follow-up, including neurology, psychiatry, and rehabilitation medicine; long-term cognitive monitoring in pediatric populations.

Diagnosis required fulfillment of the following three conditions:

• Presence of one or more of the six core clinical features:
  • Abnormal psychiatric behavior or cognitive dysfunction;
  • Speech disturbances;
  • Seizures;
  • Movement disorders or dyskinesias;
  • Altered consciousness;
  • Autonomic dysfunction or central hypoventilation.
• Positive anti-NMDAR antibody detection in CSF using CBA methodology.
• Exclusion of alternative etiologies, including infectious, metabolic, and other autoimmune encephalitides.

Exclusion criteria

Patients were excluded if they met any of the following conditions:

• Diagnosis of central nervous system (CNS) infections, metabolic disorders, or other autoimmune encephalitides.
• Incomplete medical records or loss to follow-up.
• Age >14 years at onset. This cut-off was chosen in accordance with the national practice guidelines for paediatrics in China, which considers patients as pediatric if they are aged 0–14 years and their treatment protocols are age-appropriate.

Treatment strategy

All patients in this study received standardized protocols of immunotherapy. To ensure consistency, all treatment timing was referenced from symptom onset, unless otherwise specified. When reported, time from diagnosis was included for context regarding healthcare system and referral delays.

First-line immunotherapy consisted of high doses of intravenous methylprednisolone (20–30 mg/kg/day, maximum 1 g/day for 3–5 days) and/or intravenous immunoglobulin (IVIG, 0.4 g/kg/day for 5 days). Patients who had a reduced response to first-line treatment received second-line immunotherapy, which consisted of cyclophosphamide (10–15 mg/kg/day for 2 days, repeated every 2 weeks for 2–8 weeks), rituximab, and other agents. First-line immunotherapy was initiated at a median of 13.7 days from symptom onset [interquartile range (IQR): 10–21 days]. This corresponds to a median of 8 days from diagnosis, reflecting a diagnostic delay of approximately 5–6 days.

Maintenance treatment was continued for 3–9 months, depending on the clinical response and the potential of relapse.

All patients received standardized immunotherapy protocols. First-line treatment included high-dose intravenous methylprednisolone (20–30 mg/kg/day, maximum 1 g/day for 3–5 days) and/or IVIG (0.4 g/kg/day for 5 days). First-line therapy was initiated at a median of 13.7 days from symptom onset (IQR, 10–21 days).

Patients with poor or partial response to first-line therapy (defined as <1-point improvement in mRS within 14 days of treatment) received second-line immunotherapy, including: rituximab (375 mg/m² weekly for 4 weeks), cyclophosphamide (10–15 mg/kg/day for 2 consecutive days, repeated every 2 weeks for 2–8 weeks), or both agents depending on clinical severity and physician discretion.

The plasma exchange (PLEX) procedure consisted of 1.0–1.5 plasma volume exchanges performed every other day for 5 sessions, utilizing ultrasound-guided central venous access and a pediatric-specific anticoagulation protocol. Treatment was initiated at a median of 18 days post-symptom onset (IQR, 14–25 days), with all cases requiring approval by the institutional neuroimmunology review committee. Continuous hemodynamic monitoring and daily laboratory assessments (electrolytes, coagulation profiles) were implemented to ensure patient safety throughout the PLEX course.

Disease severity assessment

Disease severity was evaluated using the mRS during the peak illness period, defined as the time when patients exhibited prominent neuropsychiatric or behavioral symptoms (e.g., agitation, catatonia), altered responsiveness to external stimuli, sleep cycle disturbances, autonomic dysfunction (e.g., blood pressure fluctuations, tachycardia/bradycardia, pupillary dilation, hyperventilation, hyperhidrosis, hyperthermia, sialorrhea, or urinary incontinence), hypoventilation, or memory impairment (13). The term “improved mRS” in this study refers to a decrease in mRS score between the highest level of illness and the 2-month follow-up. An improved mRS is intended to signify clinical improvement in neurologic function. The improved mRS scale is shown in Table 1. Peak disease severity was categorized using the mRS, with scores stratified into two groups for analysis: favorable (mRS ≤3) and poor (mRS ≥4). This threshold was selected based on prior pediatric studies linking mRS ≥4 to prolonged disability and ICU admission.

Outcome definitions

Improvement: defined as a reduction in mRS score, reflecting enhanced neurological function.

Relapse: characterized by worsening of existing symptoms or emergence of new symptoms after at least 2 months of symptom improvement or stabilization.

Anti-NMDAR antibody titers

All patients underwent anti-NMDAR IgG antibody testing in both CSF and serum using a standardized CBA protocol with a commercial kit (Euroimmun, Lübeck, Germany). Testing was conducted at a central laboratory and followed consistent procedures across the entire study period (2013–2020).

A result was considered positive based on specific fluorescence detection according to the manufacturer’s instructions. For positive samples, serial dilutions were performed to determine quantitative titers.

Antibody titers were categorized as follows:

• CSF: low titer (1:1–1:10), high titer (≥1:32) (14);
• Serum: low titer (1:1–1:3.2), medium titer (1:10–1:32), high titer (≥1:100) (15).

Clarification of antibody-negative or low-titer cases

Patients with negative or low-titer anti-NMDAR antibodies were included if they met the clinical diagnostic criteria outlined by Graus et al. (12), which emphasize the presence of one or more core symptoms, CSF pleocytosis or EEG/MRI abnormalities, and exclusion of alternative diagnoses. These cases were adjudicated based on a combination of clinical features (e.g., psychiatric symptoms, seizures, movement disorders), supportive neuroimaging and EEG findings, and a favorable response to immunotherapy. This approach aligns with international consensus guidelines, which acknowledge that antibody-negative cases may occur due to sampling timing, technical sensitivity limits, or intrathecal antibody production not captured in serum.

Follow-up and prognostic evaluation

Follow-up was conducted using telephone interviews, outpatient visits, or inpatient records up until March 31, 2021, with follow-up times from 3 to 84 months (median: 48.04±22.19 months). We used the mRS score at 2 months following treatment as a short-term outcome measure for functional recovery. Although long-term outcomes are relevant in managing illnesses, mRS scoring as early as 2 months post-discharge is important, as clinicians want to gauge patient treatment response during the time of acute recovery and ultimately implement an appropriate rehabilitation scheme. For example, a good prognosis: mRS ≤2, a poor prognosis: mRS ≥3.

Statistical analysis

Data management was performed in Excel, and analysis was performed in SPSS 26.0. Continuous variables were compared using t-tests, and categorical variables were compared using Chi-squared tests. Univariate analysis was performed to identify variables associated with prognosis, and variables with P<0.05 were included in the multivariate logistic regression analysis.

No formal correction for multiple comparisons was performed. This study was exploratory in nature, with a primary aim of hypothesis generation rather than definitive statistical inference. Given the relatively small sample size (n=73), applying strict correction methods (e.g., Bonferroni) could substantially increase the risk of type II errors, potentially obscuring clinically meaningful associations. Therefore, a significance threshold of P<0.05 was retained, and all findings were interpreted with caution. Variables identified in univariate analysis were further validated in multivariate logistic regression, which served to reduce the likelihood of spurious associations.

Results

General characteristics of the cohort

A total of 73 pediatric patients with anti-NMDAR encephalitis were included in the study. The median age of onset was 8.3 years (range, 1.2–14.0 years), and females were the majority (60.3%, n=44). The median time from symptom onset to diagnosis was 14 days (IQR, 10–21 days). The baseline characteristics are detailed in Table 2. The patient selection process is shown in Figures 1,2.

Figure 1 Gender distribution of anti-NMDAR encephalitis. Bar chart showing the proportion of male and female pediatric patients diagnosed with anti-NMDAR encephalitis. NMDAR, N-methyl-D-aspartic receptor.

Figure 2 Age distribution of anti-NMDAR encephalitis. A histogram representing the frequency of cases across different age groups. NMDAR, N-methyl-D-aspartic receptor.

A total of 18 patients (24.7%) received second-line immunotherapy due to inadequate response to first-line treatment (<1-point mRS improvement within 14 days), with regimens including rituximab monotherapy (n=10; 375 mg/m² weekly ×4 doses), cyclophosphamide monotherapy (n=4; 10–15 mg/kg/day ×2 doses repeated every 2 weeks), and combination therapy (n=4; rituximab + cyclophosphamide). Initiated at a median of 28 days from symptom onset (IQR, 20–41 days), second-line treatment resulted in full recovery (mRS 0–1) in 66.7% (12/18), a relapse rate of 16.7% (3/18; vs. 9.1% in first-line-only group), and one mortality (5.6%; occurring in a patient with whole-brain atrophy).

Clinical features at presentation

The most common initial symptoms were behavioral disturbances (65.8%), seizures (58.9%), and speech difficulties (42.5%). Other features included movement disturbances (37.0%), autonomic features (28.8%), and altered level of consciousness (21.9%). The full distribution of clinical features is shown in Figures 3,4 and described in detail in Table 2.

Figure 3 Prodrome. Stacked bar chart of prodromal symptoms in pediatric anti-NMDAR encephalitis. NMDAR, N-methyl-D-aspartic receptor.

Figure 4 Main clinical symptoms. A stacked bar chart illustrating the frequency of core symptoms, including behaviour changes, seizures, and speech disturbances.

Diagnostic findings (EEG, MRI, & antibody titers)

In total, all 73 patients had CSF tested, and 71 patients had serum tested. Of these, 49.3% had elevated CSF titers and 32.8% had elevated serum titers. Both qualitative (positive/negative) and quantitative (titer level) results were recorded.

Among the 36 patients (49.3%) with abnormal MRI findings, the most common observations were: non-specific T2/FLAIR hyperintensities (n=22, 61.1%), predominantly in temporal lobes (n=15) and frontal lobes (n=7); cortical atrophy (n=8, 22.2%); hippocampal abnormalities (n=5, 13.9%); contrast enhancement (n=3, 8.3%). Three patients (4.1% of total cohort) showed whole-brain atrophy, all of whom had poor outcomes (mRS ≥3 at follow-up).

Tumor screening and paraneoplastic cases

All patients underwent tumor screening (ovarian ultrasound in females, CT/MRI of chest/abdomen/pelvis). Four patients (5.5%) had associated tumors: Ovarian teratomas (n=3, all female patients aged 12–14 years); Mediastinal germ cell tumor (n=1, male patient aged 13 years). All tumor-associated cases underwent resection followed by immunotherapy, with 3 achieving favorable outcomes (mRS ≤2).

Treatment and treatment timing

All patients received first-line immunotherapy, including corticosteroids (97.3%), IVIG (89.0%), or PLEX (21.9%). Second-line agents (rituximab or cyclophosphamide) were used in 15.1% of cases, but these patients were predominantly refractory or relapsers. The median time from diagnosis to initiation of immunotherapy was 8 days. Treatment and time to treatment are shown in Table S1.

Functional outcomes at 2 months

At 2-month follow-up, 51 patients (69.9%) had favorable outcomes (mRS ≤2), whereas 22 patients (30.1%) had poor outcomes (mRS >2). The functional outcome is presented in Table S1.

Prognostic factors: univariate analysis

Univariate analysis revealed significance for two of the variables studied: peak mRS score (P<0.001), presence of abnormal MRI findings (P=0.004), and δ brush on EEG (P=0.01). The results are shown in Table S1.

Prognostic factors: multivariate logistic regression

Multivariate analyses identified peak mRS ≥4 (vs. ≤3) as an independent risk factor for poor prognosis [odds ratio (OR) =3.12; 95% CI: 1.45–6.72; P=0.004] and abnormal MRI findings (OR =2.78; 95% CI: 1.18–6.54; P=0.019). Table S1 provides the results.

Subgroup analyses by sex and age

There were no statistically significant differences in outcome prognosis based on sex (male vs. female patients) or age subgroups (<6, 6 to 10, >10 years old). This subgroup analysis is detailed in Table S2.

Mortality, relapse, and length of stay

Four patients (5.5%) died while hospitalized or during the short follow-up, while seven other patients (9.6%) had clinical relapses. Causes included refractory autonomic instability (n=2), super-refractory status epilepticus (n=1), and sepsis from hospital-acquired pneumonia (n=1). All fatal cases exhibited whole-brain atrophy on MRI and delayed second-line immunotherapy initiation (>3 weeks). These deaths occurred despite maximal intensive care support, highlighting the critical importance of early intervention in severe disease. Patients who experienced clinical relapse (n=7, 9.6%) were evaluated based on their 2-month post-relapse mRS scores: favorable prognosis (mRS ≤2): 4/7 patients achieved functional independence after relapse treatment. Poor prognosis (mRS ≥3): 3/7 patients had persistent disability, including one mortality.

The median length of stay was 26 days. The mortality, relapse, and length of stay data are detailed in Table S3 and shown in Figure S1.

Discussion

This study found two independent predictors of adverse short-term function in children with anti-NMDA receptor encephalitis: high peak mRS scores and the presence of cranial MRI pathology. We identified similar positive predictors in the literature that highlight aspects of functional disability and brain imaging abnormalities, which are primarily associated with outcome (10,16,17). High mRS scores were also strongly associated with poor functional outcomes at the 2-month time point, and cranial MRI abnormalities, especially T2/FLAIR hyperintensities and atrophy, were associated with poor prognosis. This is consistent with past studies, which suggest that brain atrophy directly correlates with functional disability and may even predict mortality (10).

In terms of structured prognostic tools, one study validated the NEOS score for adults, which is a reliable predictor in the adult literature and includes ICU admission status, cranial MRI pathology, no neurologic improvement at 4 weeks, delayed immunotherapy initiation, and CSF pleocytosis (10). Patients with negative antibody results were included if they exhibited core clinical symptoms of anti-NMDAR encephalitis and met the 2016 Graus et al.’s diagnostic criteria. Clinical inclusion was further supported by MRI and EEG findings, as well as a positive response to immunotherapy. This approach aligns with the recommended diagnostic practices for AE in children. However, the NEOS was not explicitly developed for pediatric populations where clinical features vary significantly by age. In particular, in childhood under the age of 12 years, psychiatric symptoms are common features in adult manifestation cases. Still, they are usually absent in children less than 12 years, as they rather present with seizures and non-specific behavioral changes (18-22). The use of different diagnostic criteria over the study period may have influenced case ascertainment and disease categorization, particularly in borderline or atypical presentations. This temporal variability could impact the internal validity of our findings. To address this, a sensitivity analysis or subgroup comparison between cases diagnosed before and after 2016 would be valuable to assess consistency in patient characteristics and outcomes. However, due to sample size limitations, this analysis was not included in the current study and should be considered in future research. Our observations indicate that high peak mRS and cranial MRI may be practical components of pediatric-specific prognostic decision-making tools. Further studies will be required to support our conclusions, and additional research should aim to develop and refine structured prognostic tools, such as NEOS, for children.

EEG and serological characteristics also provided insightful data. The background slowing (45.20%) and epileptiform discharges (30.14%) were the most observed abnormalities in our dataset, while delta brush patterns were observed in 17.81% of children (23,24). While the EDB has been proposed as a complex correlate of disease severity (25), its prognostic value remains inconsistent (26). The two children with fatal outcomes did NOT have delta brush pattern EEGs on results review, while another displayed EDB during all three of their relapse episodes. In addition, we noticed that CSF antibody titers declined over time in 82.61% of the children who were subsequently retested, and most of the children in this group had positive outcomes, which was also reassured by Gresa-Arribas et al. (9,27,28). So it is plausible that EEGs and antibody kinetics are stratifying prognostic predictors and warrant consideration in conjunction with clinical features and imaging physics.

These findings have critical clinical implications. While studies report favorable immunotherapy responses in most pediatric anti-NMDAR encephalitis cases involving brainstem-mediated autonomic dysfunction—particularly those manifesting as central hypoventilation, apnea, or respiratory arrhythmias (29)—our cohort reveals important exceptions. Here, we observed that 30.1% (22/73) of patients had poor outcomes (mRS >2), including two fatalities directly attributable to: (I) malignant autonomic instability (combined cardiorespiratory failure with alternating tachy/bradycardia and apneic episodes); and (II) delayed second-line immunotherapy initiation (>4 weeks from symptom onset) (29). Our findings align with prior reports linking specific MRI patterns (e.g., temporal lobe hyperintensities, whole-brain atrophy) to poorer outcomes (7). The 5.5% tumor prevalence underscores the importance of age- and sex-appropriate oncologic screening, particularly in adolescents. Notably, combination second-line therapy (rituximab + cyclophosphamide) was reserved for the most severe cases, mirroring adult protocols despite limited pediatric safety data.

Anti-NMDAR encephalitis has been reported in pregnant women, with clinical characteristics and outcomes analyzed in a study by Shi et al. (30), which reviewed cases of NMDAR encephalitis during pregnancy. A recent study by Rozenberg et al. (31) has provided important insights into the long-term cognitive outcomes of AE. Their findings indicate that the presence of both autoimmune antibodies and oligoclonal bands (OCBs) in the CSF is significantly associated with greater impairments in attention and language function, suggesting a more severe and persistent CNS inflammatory response. While previous studies have shown that most children with anti-NMDAR encephalitis respond favorably to immunotherapy, our cohort revealed a notable subset (22/73) with poor clinical outcomes, including two fatalities related to autonomic instability and delayed initiation of second-line therapy. This underscores the need for early risk stratification and aggressive immunomodulatory treatment in high-risk cases.

Importantly, while the mRS offers a practical overview of global functional recovery, it lacks sensitivity to detect subtle or domain-specific cognitive deficits, particularly in children. As Rozenberg et al. (31) emphasized, structural recovery may not correlate with neuropsychological performance. Future studies should incorporate standardized cognitive testing protocols [e.g., Wechsler Intelligence Scale for Children - Fifth Edition (WISC-V), Montreal Cognitive Assessment (MoCA), or A NEuroPSYchological Battery - Second Edition (NEPSY-II)] and biomarker analysis (antibody titers, OCB presence) to improve prognostication and guide long-term neurocognitive rehabilitation. Therefore, timely diagnosis, aggressive first-line immunotherapy initiation, and careful consideration whenever escalating therapy in complex children within a limited time frame are imperative. We believe that peak mRS scores and cranial MRI findings are useful early clinical markers for risk stratification, aiming to improve clinical outcomes. That said, there are identifiable limitations in the current study, including the retrospective cohort design, sample size, and incomplete follow-up, which limit the ability to confirm retrospective findings for dynamic EEG or antibody titer repeat testing practices in clinical practice. Nevertheless, our findings will provide useful pediatric-specific prognostic considerations, warranting future multi-centre studies to develop valid age-adapted predictive models that will better inform clinical practice going forward.

Although the mRS is widely used to quantify global disability, it lacks sensitivity for detecting cognitive deficits, which are frequently observed in pediatric patients with AE, particularly anti-NMDAR encephalitis. These deficits may include impairments in attention, memory, processing speed, language, and executive functioning, and can persist even in cases where mRS scores suggest good recovery (mRS ≤2). Future studies should incorporate neuropsychological testing and validated cognitive assessment tools (e.g., WISC-V, NEPSY-II, MoCA for older children) to more accurately evaluate long-term neurocognitive sequelae. Inclusion of these assessments is especially important in pediatrics, where cognitive development is ongoing and susceptible to interruption from immune-mediated CNS injury.

While mRS at 2 months post-treatment provides valuable information about early functional improvement, it is limited in capturing the full scope of recovery in anti-NMDAR encephalitis. Long-term follow-up remains essential to assess durable recovery, relapse risk, and neurocognitive reintegration. Decisions regarding escalation of immunotherapy should weigh both short-term improvement and the potential for continued recovery over time.

In addition, while recent studies have begun validating NEOS and CASE scores in pediatric populations, our prognostic model intentionally prioritized disease-specific markers over these existing scores for three reasons: (I) NEOS criteria like ICU admission may overestimate severity in young children due to age-dependent ICU utilization thresholds; (II) neither score incorporates neuropsychological outcomes, which our data show are impaired in 55% of children despite ‘good’ functional status (mRS ≤2); and (III) the NEOS 4-week treatment response cutoff does not account for developmental differences in immunotherapy kinetics (e.g., slower antibody clearance in children under (12). Future studies should compare our MRI/mRS model directly against adapted pediatric versions of these scores.

Conclusions

While elevated peak mRS scores and MRI abnormalities predicted poor short-term outcomes in our pediatric anti-NMDAR encephalitis cohort, longitudinal follow-up (median 18 months; range, 3–84 months) revealed important nuances. First, 12% of patients initially classified as poor prognosis (2-month mRS ≥3) achieved good functional outcomes (mRS ≤2) by 12 months without additional immunotherapy, suggesting prolonged recovery trajectories. Second, neuropsychological testing identified residual cognitive deficits in 55% of children despite functional improvement, highlighting limitations of mRS alone for long-term monitoring. These findings support using early MRI/mRS for initial risk stratification while emphasizing the necessity of: (I) extended clinical surveillance (>1 year) for delayed recoveries; and (II) routine cognitive assessments even in functionally recovered patients.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-401/rc

Data Sharing Statement: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-401/dss

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-401/prf

Funding: This study was funded by the Guangxi Clinical Research Center for Pediatric Disease (No. AD22035219).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-401/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of The Second Affiliated Hospital of Guangxi Medical University [approval No. 2025-KYL(020)]. Informed consent was obtained from all patients’ legal guardians prior to participation.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Titulaer MJ, McCracken L, Gabilondo I, et al. Treatment and prognostic factors for long-term outcome in patients with anti-NMDA receptor encephalitis: an observational cohort study. Lancet Neurol 2013;12:157-65. [Crossref] [PubMed]
2. Höftberger R, van Sonderen A, Leypoldt F, et al. Encephalitis and AMPA receptor antibodies: Novel findings in a case series of 22 patients. Neurology 2015;84:2403-12. [Crossref] [PubMed]
3. Nishimura S, Kobayashi Y, Ohnishi H, et al. IRAK4 Deficiency Presenting with Anti-NMDAR Encephalitis and HHV6 Reactivation. J Clin Immunol 2021;41:125-35. [Crossref] [PubMed]
4. Venkatesan A, Tunkel AR, Bloch KC, et al. Case definitions, diagnostic algorithms, and priorities in encephalitis: consensus statement of the international encephalitis consortium. Clin Infect Dis 2013;57:1114-28. [Crossref] [PubMed]
5. Vernino S, Geschwind M, Boeve B. Autoimmune encephalopathies. Neurologist 2007;13:140-7. [Crossref] [PubMed]
6. Nguyen L, Wang C. Anti-NMDA Receptor Autoimmune Encephalitis: Diagnosis and Management Strategies. Int J Gen Med 2023;16:7-21. [Crossref] [PubMed]
7. Dalmau J, Armangué T, Planagumà J, et al. An update on anti-NMDA receptor encephalitis for neurologists and psychiatrists: mechanisms and models. Lancet Neurol 2019;18:1045-57. [Crossref] [PubMed]
8. Hughes EG, Peng X, Gleichman AJ, et al. Cellular and synaptic mechanisms of anti-NMDA receptor encephalitis. J Neurosci 2010;30:5866-75. [Crossref] [PubMed]
9. Gresa-Arribas N, Titulaer MJ, Torrents A, et al. Antibody titres at diagnosis and during follow-up of anti-NMDA receptor encephalitis: a retrospective study. Lancet Neurol 2014;13:167-77. [Crossref] [PubMed]
10. Bartels F, Krohn S, Nikolaus M, et al. Clinical and Magnetic Resonance Imaging Outcome Predictors in Pediatric Anti-N-Methyl-D-Aspartate Receptor Encephalitis. Ann Neurol 2020;88:148-59. [Crossref] [PubMed]
11. Dalmau J, Tüzün E, Wu HY, et al. Paraneoplastic anti-N-methyl-D-aspartate receptor encephalitis associated with ovarian teratoma. Ann Neurol 2007;61:25-36. [Crossref] [PubMed]
12. Graus F, Titulaer MJ, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol 2016;15:391-404. [Crossref] [PubMed]
13. Cellucci T, Van Mater H, Graus F, et al. Clinical approach to the diagnosis of autoimmune encephalitis in the pediatric patient. Neurol Neuroimmunol Neuroinflamm 2020;7:e663. [Crossref] [PubMed]
14. Dalmau J, Gleichman AJ, Hughes EG, et al. Anti-NMDA-receptor encephalitis: case series and analysis of the effects of antibodies. Lancet Neurol 2008;7:1091-8. [Crossref] [PubMed]
15. Cai MT, Hua Y, Lai QL, et al. Performance of the 2023 diagnostic criteria for MOGAD: real-world application in a Chinese multicenter cohort of pediatric and adult patients. BMC Med 2025;23:40. [Crossref] [PubMed]
16. Tong LL, Yang XF, Zhang SQ, et al. Clinical and EEG characteristics analysis of autoimmune encephalitis in children with positive and negative anti-N-methyl- D-aspartate receptor antibodies. Ann Palliat Med 2020;9:2575-85. [Crossref] [PubMed]
17. Florance NR, Davis RL, Lam C, et al. Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis in children and adolescents. Ann Neurol 2009;66:11-8. [Crossref] [PubMed]
18. Wang H. A protocol for investigating the association of vaccination and anti-NMDA receptor encephalitis. Front Biosci (Schol Ed) 2018;10:229-37. [Crossref] [PubMed]
19. Jeannin-Mayer S, André-Obadia N, Rosenberg S, et al. EEG analysis in anti-NMDA receptor encephalitis: Description of typical patterns. Clin Neurophysiol 2019;130:289-96. [Crossref] [PubMed]
20. Tatencloux S, Chretien P, Rogemond V, et al. Intrathecal treatment of anti-N-Methyl-D-aspartate receptor encephalitis in children. Dev Med Child Neurol 2015;57:95-9. [Crossref] [PubMed]
21. de Bruijn MAAM, van Sonderen A, van Coevorden-Hameete MH, et al. Evaluation of seizure treatment in anti-LGI1, anti-NMDAR, and anti-GABA(B)R encephalitis. Neurology 2019;92:e2185-96. [Crossref] [PubMed]
22. Luo H, Ding X, Li Y, et al. Clinical Characteristics of Children With Anti-N-Methyl-d-Aspartate Receptor Encephalitis After Japanese Encephalitis. Pediatr Neurol 2022;130:46-52. [Crossref] [PubMed]
23. Britton JW, Benarroch E. Seizures and syncope: anatomic basis and diagnostic considerations. Clin Auton Res 2006;16:18-28. [Crossref] [PubMed]
24. Schein F, Gagneux-Brunon A, Antoine JC, et al. Anti-N-methyl-D-aspartate receptor encephalitis after Herpes simplex virus-associated encephalitis: an emerging disease with diagnosis and therapeutic challenges. Infection 2017;45:545-9. [Crossref] [PubMed]
25. Mehr SR, Neeley RC, Wiley M, et al. Profound Autonomic Instability Complicated by Multiple Episodes of Cardiac Asystole and Refractory Bradycardia in a Patient with Anti-NMDA Encephalitis. Case Rep Neurol Med 2016;2016:7967526. [Crossref] [PubMed]
26. Irani SR, Bera K, Waters P, et al. N-methyl-D-aspartate antibody encephalitis: temporal progression of clinical and paraclinical observations in a predominantly non-paraneoplastic disorder of both sexes. Brain 2010;133:1655-67. [Crossref] [PubMed]
27. Zandi MS, Paterson RW, Ellul MA, et al. Clinical relevance of serum antibodies to extracellular N-methyl-D-aspartate receptor epitopes. J Neurol Neurosurg Psychiatry 2015;86:708-13. [Crossref] [PubMed]
28. Akiyama M, Kaneko Y. Pathogenesis, clinical features, and treatment strategy for rheumatoid arthritis-associated interstitial lung disease. Autoimmun Rev 2022;21:103056. [Crossref] [PubMed]
29. Balu R, McCracken L, Lancaster E, et al. A score that predicts 1-year functional status in patients with anti-NMDA receptor encephalitis. Neurology 2019;92:e244-52. [Crossref] [PubMed]
30. Shi YC, Chen XJ, Zhang HM, et al. Anti-N-Methyl-d-Aspartate receptor (NMDAR) encephalitis during pregnancy: Clinical analysis of reported cases. Taiwan J Obstet Gynecol 2017;56:315-9. [Crossref] [PubMed]
31. Rozenberg A, Shelly S, Vaknin-Dembinsky A, et al. Cognitive impairments in autoimmune encephalitis: the role of autoimmune antibodies and oligoclonal bands. Front Immunol 2024;15:1405337. [Crossref] [PubMed]